{"schema_version":"papers/paper-detail-v1","title":"How Much Online RL is Enough? Informative Rollouts for Offline Preference Optimization in RLVR","surface":"papers","opportunity_kernel":{"paper_id":"9951b577-6f3b-40cc-91d7-3add1cc34246","title":"How Much Online RL is Enough? Informative Rollouts for Offline Preference Optimization in RLVR","authors":["Richa Verma","Balaraman Ravindran"],"arxiv_id":"2605.21266v1","doi":null,"published_at":"2026-05-20T14:53:43.000Z","score_object":{"overall":{"value":7,"scale":"0-10","confidence":0.85,"confidence_reason":"Backfilled from persisted papers.viability_score.","model_version":"phase0-backfill-v1","pipeline_version":"phase0-kernel-v1","computed_at":"2026-05-21T20:10:39.773Z","fresh_until":"2026-06-20T20:10:39.773Z","is_stale":false,"source_count":1,"missingness":[]},"technical":{"value":1.4,"scale":"0-10","confidence":0.55,"confidence_reason":"Backfilled from paper_extraction_scorecards.reconstruction_score.","model_version":"phase0-backfill-v1","pipeline_version":"phase0-kernel-v1","computed_at":"2026-05-21T20:13:55.064Z","fresh_until":"2026-06-04T20:13:55.064Z","is_stale":false,"source_count":1,"missingness":["reproducibility_results.reproducibility_score","deployability_scores.score"]},"commercial":{"value":4,"scale":"0-10","confidence":0.75,"confidence_reason":"Backfilled from persisted commercial_flags and repo availability.","model_version":"phase0-backfill-v1","pipeline_version":"phase0-kernel-v1","computed_at":"2026-05-21T20:10:39.773Z","fresh_until":"2026-06-20T20:10:39.773Z","is_stale":false,"source_count":1,"missingness":[]},"market":{"value":2,"scale":"0-10","confidence":0.45,"confidence_reason":"Heuristic market score materialized from paper metadata, repo availability, and deployability.","model_version":"phase0-backfill-v1","pipeline_version":"phase0-kernel-v1","computed_at":"2026-05-21T20:28:20.052Z","fresh_until":"2026-06-04T20:28:20.052Z","is_stale":false,"source_count":1,"missingness":[]},"team":{"value":4,"scale":"0-10","confidence":0.42,"confidence_reason":"Heuristic fallback from paper author count and extracted affiliations.","model_version":"phase0-backfill-v1","pipeline_version":"phase0-kernel-v1","computed_at":"2026-05-21T20:10:39.773Z","fresh_until":"2026-06-04T20:10:39.773Z","is_stale":false,"source_count":2,"missingness":[]},"methodology":{"value":3.9,"scale":"0-10","confidence":0.82,"confidence_reason":"Backfilled from paper_extraction_scorecards.total_score.","model_version":"phase0-backfill-v1","pipeline_version":"phase0-kernel-v1","computed_at":"2026-05-21T20:13:55.064Z","fresh_until":"2026-06-20T20:13:55.064Z","is_stale":false,"source_count":2,"missingness":[]}},"evidence_receipt":{"freshness":"stale","proof_status":"unverified","repo_status":"missing","references_count":0,"source_count":3,"coverage":0.5,"missingness":["repo_url","references","proof_status"],"unresolved_unknowns":["proof verification has not been recorded yet"],"last_verification_at":"2026-05-21T20:28:20.052Z"},"lineage_hash":"3fc65f0f4e1730a24dd731394f838cac826c40a35638d0d83033ce2e9cbbc2a3"},"distribution":null,"replication_evidence":[],"author_dna":[]}